|
Currently, Jiali Chen is the second-year direct Ph.D. student at Key Laboratory of Big Data and Intelligent Robot of South China University of Technology (SCUT), supervised by Prof. Yi Cai. He works closely with Dr. Jiayuan Xie at Hong Kong Polytechnic University (PolyU). Before that, He also obtained the B.E. degree in Department of Software Engineering from South China University of Technology (SCUT) in 2023. His research interests revolve around Multimodal Reasoning, Causal Inference and Vision & Language. Feel free to contact me if you're interested in discussing or seeking potential collaborations. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
CADReview: Automatically Reviewing CAD Programs with Error Detection and Correction
Jiali Chen*, Xusen Hei*, Hongfei Liu, Yuancheng Wei, Zikun Deng, Jiayuan Xie, Yi Cai, Qing Li Main Conference of Annual Meeting of the Association for Computational Linguistics, ACL 2025 [Arxiv], [Code], [Project Page] Area: CAD Modeling, MLLMs We propose ReCAD, a framework that detects and corrects errors in CAD programs by aligning 3D object designs with reference images. Our newly introduced CADReview dataset and experiments show ReCAD significantly outperforms existing MLLMs in CAD error correction. |
|
Classic4Children: Adapting Chinese Literary Classics for Children with Large Language Model
Jiali Chen, Xusen Hei, Yuqi Xue, Zihan Wu, Jiayuan Xie, Yi Cai Findings the Nations of the Americas Chapter of the ACL, NAACL 2025 [Paperlink], [Code] Area: Large Language Model, Text Style We highlight childrenβs reading preferences: vivid character portrayals, concise narrative structure and appropriate readability are essential in adapting Chinese literary classics for children. Our proposed InstructChild explicitly leverages these preferences to guide the LLM in generating child-friendly text for children. Additionally, we construct the Classic4Children dataset for a comprehensive evaluation. |
|
Learning to Correction: Explainable Feedback Generation for Visual Commonsense Reasoning Distractor
Jiali Chen, Xusen Hei, Yuqi Xue, Yuancheng Wei, Jiayuan Xie, Yi Cai, Qing Li ACM Multimedia, ACM MM 2024 [Paperlink], [Code] Area: Large Multimodal Model, New Benchmark We present the work to investigate the error correction capabilities of large multimodal models (LMMs), construct a new benchmark and introduce the feedback generation task for evaluation. I would like to extend my heartfelt gratitude to my girlfriend, Ms. Wen, for inspiring the idea behind this paper. |
|
Deconfounded Emotion Guidance Sticker Selection with Causal Inference
Jiali Chen, Yi Cai, Ruohang Xu, Jiexin Wang, Jiayuan Xie, Qing Li ACM Multimedia, ACM MM 2024 [Paperlink] Area: Bias, Causal Inference, Sticker Selection This paper presents a Causal Knowledge-Enhanced Sticker Selection model that addresses spurious correlations in sticker selection by using a causal graph and a knowledge-enhanced approach. |
|
Knowledge-Augmented Visual Question Answering with Natural Language Explanation
IEEE Transactions on Image Processing, TIP 2024 [Paperlink], [Code] Area: VQA, Multimodal Reasoning We introduce KICNLE, which generates consistent answer and explanation with external knowledge. |
|
Deconfounded Visual Question Generation with Causal Inference
Jiali Chen, Zhenjun Guo, Jiayuan Xie, Yi Cai, Qing Li ACM Multimedia, ACM MM 2023 [Paperlink], [Code] Area: Bias, Causal Inference, Visual Question Generation We identify previous models frequently learn highly co-occurring object relationships and attributes, which is an inherent bias in question generation. This study first introduces a causal perspective on VQG and adopts the causal graph to analyze spurious correlations among variables. We propose KECVQG mitigates the impact of spurious correlations for VQG. |
|
|
|
|
|
Last updated on May, 2025
This website template borrowed from [here]
|